> You seem to be in agreement with Hamp about the development of *-
o:u (in
> masculine *dwo:u) into *-u: > Alb. -y. But if /-y/ is a "regular"
> masculine dual endings, the simplest derivation of feminine /-y:/
from
> it is preferable. *dwoi(h1)- + -o:u unnecessarily complicates
things.
> *-oi- normally gives Albanian /e/, and you need more ad
hoc "rules" to
> explain the modern vowel. Modern Albanian length is not inherited
but
> results from relatively late syllable contractions. How do you
propose
> to get /y:/ from expected /e + y/ in post-PAlb. times?
Hello Piotr,
My main assumption is that (and I hardly tried to respect this
rule):
"The single source of Albanian y is iu (u:) or ui (knowing also
that the intermediate stage of u: was iu (my opinion)"
Note: Other opinions (like Orel) proposed ui as intermediate
stage of u: (but this not change the assumption above).
I know that different authors have proposed other sources for
Albanian y, but their examples are not very reliables.
If you know a good example showing a different source of
Albanian y please to indicate it here...
a) So in my opinion o:u cannot generate y (as Hamp proposed). And
before this o:u cannot generates u: also ...
b) Also (sorry but I couldn't find another example to can accept
your rule) I have also doubts that *ú[w]o: (in dú[w]o:) can generate
u: (via *úö:). Please to post another example to can validate your
proposal.
Next trying to respect the rule above both sy and dy needs an u:
(and my issue is that I cannot identified well the source of this
u:) I said this because both PIE Roots already contains the
necessary i: but there is no trace of u:
From here my derivations:
a) PIE *dwo-ih1 + uh1 > PAlb *dwai:u: > [ai >e] PAlb *dwei:u: >
PAlb [w>zero; ei>i] > dju: > [iu:>y] > Alb. dy 'two'
The derivation is ok for me (there is no 'ad-hoc' rule or '/e+y/'
as you told me, due to the right timeframe of each transformation:
1. ai > e
2. ei > i
3. (i)u: > y
b) PIE *h3e/okW-ih1 + uh1 [h3e/o > o >a ; kW/i>c^] > PAlb *ac^i:-u:
> [c^> c > s] PAlb *asi:u: > [a >zero ; i:u: > y(:)]> Alb. sy(:)'eye'
The issue that remains is 'from where this u:' ? But whatever was
its source ( :) ) if 'the single source of y is iu (u:) or ui ' for
sure the presence of u: is necessary.
If we will not go 'for this 'additional' u:' we are obliged to
find a reliable source for y (so a new rule) other than iu(u:) or ui.
From here 'my preference' for the 'additional u:'...
Best Regards,
Marius